Google is harder to replace because to replace Google, you actually have to replace Google. Develop their search bots, scale up your operation to their scales, sell ads in a market Google dominated, etc. Building a "better search engine" merely gets you a better search engine, you have to monetize it all the way to actually replace Google. This is intrinsically a multi-billion-dollar effort, competing with them the whole way. Distributed internet search has not been shown to be practical, I'm not sure it is even in theory.
Facebook on the other hand can be replaced with open standards and tons of little nodes everywhere. It isn't even all that hard, coordination is the biggest problem. (Or if you prefer, yeah, it's hard, but the original timeline was 20 years. It's not 20 years worth of hard. Again, think back to 1990. Consider the history of open source software since then. Creating distributed-Facebook is not harder than KDE or the Linux Kernel, each of which is only a vanishing fraction of the open source world's output in the last 20 years.)
Semantic Web standards will make Google obsolete? You haven't noticed them leading the charge to actually get these things out into the wild? The little tables of contents, phone numbers, etc in search results? And they're the best in the world, from what I can see, at semanticizing content that wasn't semantically labeled by the originator.
Facebook on the other hand can be replaced with open standards and tons of little nodes everywhere. It isn't even all that hard, coordination is the biggest problem. (Or if you prefer, yeah, it's hard, but the original timeline was 20 years. It's not 20 years worth of hard. Again, think back to 1990. Consider the history of open source software since then. Creating distributed-Facebook is not harder than KDE or the Linux Kernel, each of which is only a vanishing fraction of the open source world's output in the last 20 years.)
Semantic Web standards will make Google obsolete? You haven't noticed them leading the charge to actually get these things out into the wild? The little tables of contents, phone numbers, etc in search results? And they're the best in the world, from what I can see, at semanticizing content that wasn't semantically labeled by the originator.